REVIEWS

Non-small-cell lung cancers:

a heterogeneous set of diseases
Zhao Chen1*, Christine M. Fillmore2,3,4*, Peter S. Hammerman1, Carla F. Kim2,3,4 and
Kwok-Kin Wong1,5,6
Abstract | Non-small-cell lung cancers (NSCLCs), the most common lung cancers, are known
to have diverse pathological features. During the past decade, in-depth analyses of lung
cancer genomes and signalling pathways have further defined NSCLCs as a group of distinct
diseases with genetic and cellular heterogeneity. Consequently, an impressive list of
potential therapeutic targets was unveiled, drastically altering the clinical evaluation and
treatment of patients. Many targeted therapies have been developed with compelling
clinical proofs of concept; however, treatment responses are typically short-lived. Further
studies of the tumour microenvironment have uncovered new possible avenues to control
this deadly disease, including immunotherapy.

1
Department of Medical
Oncology, Dana-Farber
Cancer Institute, Boston,
Massachusetts 02115, USA.
2
Stem Cell Program, Boston
Children’s Hospital, Boston,
Massachusetts 02115, USA.
3
Harvard Stem Cell Institute,
Cambridge, Massachusetts
02138, USA.
4
Department of Genetics,
Harvard Medical School,
Boston, Massachusetts
02115, USA.
5
Department of Medicine,
Harvard Medical School,
Boston, Massachusetts
02115, USA.
6
Belfer Institute for Applied
Cancer Science, Dana-Farber
Cancer Institute, Boston,
Massachusetts 02115, USA.
*These authors contributed
equally to this work.
Correspondence to P.S.H.,
C.F.K. and K.-K.W.
e‑mails: phammerman@
partners.org; carla.kim@
childrens.harvard.edu;
kwong1@partners.org
doi:10.1038/nrc3775
Lung cancer results in the largest number of cancer-
related deaths worldwide1,2. More than 85% of those
cases are currently classified as non-small-cell lung
cancer (NSCLC), for which the predicted 5‑year sur-
vival rate is 15.9% — a figure that has only marginally
improved during the past few decades3. Technological
advances during the past decade, including the intro-
duction of next-generation sequencing (NGS), the
generation of multiple genetically engineered mouse
models (GEMMs) of lung cancer and the construction
of large databases characterizing the molecular fea-
tures of human tumours, have transformed our view of
NSCLC from histopathological descriptions to precise
molecular and genetic identities that can be resolved to
the single-cell level. In parallel, approaches and concepts
from fields such as developmental biology, stem cell biol-
ogy and immunology have deepened our knowledge of
tumour development, cellular heterogeneity and inter-
actions between the lung tumour and its surrounding
microenvironment. These multidisciplinary efforts
have enhanced our understanding of molecular disease
mechanisms, thereby forming the rationales for target-
ing different cellular compartments simultaneously.
Scientists and physicians have better tools than ever to
pursue answers to two provocative questions: first, how
can we define the specific subsets of NSCLC that differ
by cellular and molecular composition? Second, how
can we effectively control lung cancer growth for each
specific subset of NSCLC? In this Review, we discuss
how data that are derived from technological advances
in lung cancer genomics, mouse modelling of cancers
and tumour microenvironment studies might be used
to improve the survival of patients with NSCLC through
the development of novel therapeutic strategies.
Defining NSCLC subsets
NSCLC is currently defined by pathological character­
istics. The two predominant NSCLC histological pheno-
types are adenocarcinoma (ADC; ~50%) and squamous
cell carcinoma (SCC; ~40%)4,5. In general, ADCs arise in
more distal airways, whereas SCCs arise in more proxi-
mal airways and are more strongly associated with smok-
ing and chronic inflammation than ADCs4,5. ADCs often
have glandular histology and express biomarkers that
are consistent with an origin in the distal lung, includ-
ing thyroid transcription factor 1 (TTF1; also known
as NKX2‑1) and keratin 7 (KRT7)4,5. By contrast, SCCs
are characterized by squamous differentiation, which is
more reminiscent of the pseudostratified columnar epi-
thelium that lines the trachea and upper airways4,6. SCCs
are distinguished from ADCs in the clinic by immuno­
staining for cytokeratin 5 and cyto­keratin 6 and/or
the transcription factors SRY-box 2 (SOX2) and p63
(REFS 4,5,7). Other subtypes of NSCLC include large cell
carcinoma, which is diagnosed by exclusion if tumour
cells do not appear glandular or squamous in shape or
express ADC or SCC biomarkers, although it is unclear
whether large cell carcinomas are genetically distinct
from ADC or SCC4. Some neuroendocrine tumours
are also classified as NSCLC, although the most aggres-
sive form of neuroendocrine tumour is classified as
small-cell lung cancer (SCLC)4.

NATURE REVIEWS | CANCER VOLUME 14 | AUGUST 2014 | 535

© 2014 Macmillan Publishers Limited. All rights reserved

REVIEWS
Myeloid-derived
suppressor cells
(MDSCs). MDSCs encompass a
heterogeneous population of
myeloid cells, which share the
ability to suppress T cells
through the production of
arginase and the expression
of inducible nitric oxide
synthase (iNOS).
536 | AUGUST 2014 | VOLUME 14 www.nature.com/reviews/cancer
Genetic mutations and genomic heterogeneity.
Although histological features and marker expression
remain the basis of clinical tumour diagnosis, recent
advances in NGS and other high-throughput genomic
profiling platforms have allowed researchers to exam-
ine the breadth of genetic mutations within lung
tumours. Following the identification of KRAS and
BRAF mutations8,9, epidermal growth factor receptor
(EGFR) mutations were discovered in patients with
lung ADC and were associated with response to EGFR
inhibitors10–13. Further recurrent mutations and ampli-
fications in many potentially targetable oncogenes have
since been identified in lung ADC, including HER2
(also known as ERBB2), MET, fibroblast growth fac-
tor receptor 1 (FGFR1) and FGFR2, as well as fusion
oncogenes involving anaplastic lymphoma kinase
(ALK), the ROS1 receptor tyrosine kinase, neuregu-
lin 1 (NRG1), neurotrophic tyrosine kinase recep-
tor type 1 (NTRK1) and RET 14–22. These oncogenic
changes, many of which predict sensitivity to clini-
cal inhibitors, jointly account for most cases of lung
ADC23–25. For lung SCC, the number of tumours for
which whole-exome sequencing is available is lower
than for ADC but, so far, potentially targetable muta-
tions in ADC do not seem to be prevalent in this
histological subtype20. Instead, genes such as discoi-
din domain-containing receptor 2 (DDR2), FGFR1,
FGFR2, FGFR3 and genes in the PI3K pathway seem
to be more commonly mutated in lung SCC20. Many
of these mutations (with the exception of those in
the PI3K pathway) have been validated by preclinical
studies as driver mutations22,26,27.
NGS studies have also revealed the molecular tax-
onomy of lung cancer and have shown a dazzling com-
plexity of somatic alterations in NSCLCs that extends
far beyond protein kinases to include epigenome
modifiers, transcription factors, splicing factors and
genes involved in cellular immunity 20,28,29. Potentially
important mutations and copy number gains identified
from patient tumours are summarized in TABLE 1, with
relevant preclinical and clinical evidence. Among the
21 different tumour types for which exome sequences
were directly compared, lung SCC and ADC ranked
second and third highest in median somatic muta-
tion frequency, with an average of ten mutations per
megabase of coding DNA sequenced30. It is worth
noting that ADCs in non-smokers have 5–6‑times
fewer mutations24,31. Given this relatively large num-
ber of mutations per tumour, there will probably be
more important mutations identified for NSCLC as
the number of tumours that are analysed increases.
An important challenge that remains is understanding
which of these many mutations are important in lung
carcinogenesis and/or treatment response, in contrast
to those mutations that are merely a consequence of
the tumorigenic process. Thus, the genomic profiles
highlight the heterogeneity of the NSCLC genome and
provide a plausible explanation for the highly hetero-
geneous treatment responses that we have observed in
the clinic. By cataloguing a large collection of muta-
tions for each patient, a more accurate evaluation of
© 2014 Macmillan Publishers Limited. All rights reserved
the net effects of genotype and therapy response may be
achieved and will ultimately inform the most suitable
treatment strategies.
Other novel technologies have also facilitated the
discovery and validation of somatic mutations in lung
cancer. For example, high-throughput screens using
established short hairpin RNA (shRNA) libraries have
identified genes that cause synthetic lethality with
common oncogenic events, such as KRAS-activating
mutations or p53 inactivation, leading to potential
new treatment targets, such as TANK-binding kinase 1
(TBK1)32. Similarly, the application of mass spectro­
metry to metabolomic, proteomic and phosphoki-
nase profiling, as well as single cell time-of-flight mass
cytometry (cyTOF), have led to numerous new findings,
including the discovery of recurrent aberrations such as
the ROS1 fusions and the potential diagnostic or prog-
nostic marker isocitrate dehydrogenase 1 (IDH1)17,33,34.
Such advances in high-throughput technology are pro-
moting rapid advances in our understanding of NSCLC
biology and, ultimately, will help to determine how
NSCLC develops, spreads and can be better treated.
Heterogeneity in lung tumour microenvironments.
The concept of tumour heterogeneity applies not only
to tumour epithelial cells but also to the diverse micro-
environments with which the tumour cells interact 35.
Carcinoma cells, in the lung and other organs, are closely
associated with the extracellular matrix (ECM), mesen-
chymal cells such as fibroblasts, infiltrating immune cells
and vasculature (FIG. 1). In some cases, this environment
is essential to tumour initiation or tumour growth,
whereas in other cases it can prevent tumorigenesis or
even promote tumour clearance35,36.
In lung tumorigenesis, genesis of new blood and lym-
phatic vessels supplies necessary nutrients for tumour
growth and allows for an influx of immune cells of the
myeloid and lymphoid lineages. The myeloid cells that
are implicated in this process include tumour-associated
macrophages (TAMs) and tumour-associated neutro-
phils37. Mice that harboured germline knock‑in of kinase-
dead inhibitor of nuclear factor-κB (NF‑κB) kinase
subunit-α (IKKα) developed spontaneous lung SCC
that is characterized by NF‑κB activation and marked
accumulation of TAMs that were essential for disease
progression38. Secretion of pro-angiogenic factors such
as platelet-derived growth factor (PDGF) and vascular
endothelial growth factor (VEGF) by TAMs in lung can-
cer suggests why these cells are associated with increased
microvessel formation39,40. Likewise, increased neutro-
phil numbers have been associated with poor prognosis
in NSCLCs, perhaps owing to their ability to degrade
matrices with elastase41,42. Neutrophils that are found
in mouse tumours are phenotypically characterized as
polymorphonuclear CD11B- and lymphocyte antigen
6G‑expressing (CD11B+Ly6G+) cells, and are often con-
sidered to be a subtype of myeloid-derived suppressor cells
(MDSCs)43. In the tumour microenvironment, accu-
mulated MDSCs are thought to promote tumour pro-
gression by increasing matrix degradation, tumour cell
proliferation, metastasis and angiogenesis35,37.
 REVIEWS

Table 1 | Potential important alterations in ADC and SCC

Gene Status Frequency (%) Available GEMMs Currently Selected potential Refs
(M, C available targeted targeted therapies
or F)* ADC SCC therapies Preclinical Clinical
evidence evidence
Receptor tyrosine kinases
EGFR M or C 10 (M) 2–3 L858R, Del19, T790M and Erlotinib, gefitinib AZD9291, CO‑1686 and 126 11
Ins20 and afatinib HM61713
FGFR1 C N/A 20 N/A N/A Dovitinib, ponatinib, AZD4547 150 22
and BGJ398
FGFR2 M or C 3 (M) 3 N/A N/A Dovitinib, ponatinib, AZD4547 151 20
and BGJ398
ALK F 3–5 <1 ALK fusion, L1196M and Crizotinib and AP26113, alectinib, 125 18
F1174L ceritinib ganetespib and PF‑06463922
MET C 2–4 N/A Overexpression Crizotinib Tivantinib, cabozantinib, 152 14
INC280 and onartuzumab
ROS1 F 1–2 N/A N/A Crizotinib PF‑06463922 153 17
NTRK1 F 1–2 N/A N/A N/A Crizotinib and lestaurtinib 21 21
RET F 1 N/A N/A N/A Carbozantinib and vandetanib 154 16
HER2 M or C 2–4 N/A HER2-YVMA insertion N/A Neratinib, afatinib, lapatinib 155 19
(M) and trastuzumab
DDR2 M N/A 2–3 N/A N/A Dasatinib 27 27
PDGFRA M 6–7 4 N/A N/A Sunitinib 156 28
Signalling
KRAS M 15–25 1–2 G12D, G12C and G12V N/A Selumetinib plus docetaxel 157 158
combination
NF1 M 12 10 Null N/A 159 28
BRAF M 1–6 4–5 V600E N/A Vemurafenib, dabrafenib and N/A 160
trametinib
PIK3CA M 5 15 p110α N/A BEZ235, BKM120 and 99 161
GDC0941
MEK1 M 1 N/A N/A N/A Selumetinib and trametinib N/A 162
NOTCH1 M 8 1 Conditional null N/A N/A 163 164
Epigenetic factors
MLL2 M 9 20 N/A N/A N/A 165 28
EZH2 M 2 2 N/A N/A N/A 166 28
TET2 M 3 2 N/A N/A N/A 167 28
DNMT3A M 4 1 N/A N/A N/A 168 28
Transcription factors
SOX2 C 6 65 Overexpression N/A N/A 7 103
MYC C 25 N/A Overexpression N/A N/A 133 104
Proteolysis
KEAP1 M 17 12 N/A N/A N/A 169 170
Cell cycle
CDKN2A M 7 15 Null N/A N/A 171 172
Ligand
NRG1 F <1 N/A N/A N/A N/A 15 15
Tumour suppressor
TP53 M 52 79 Conditional null and R172H N/A N/A 98 173
LKB1 M 9 2 Conditional null N/A N/A 65 174
PTEN M 2 8 Conditional null N/A BEZ235, BKM120 and 175 176
GDC0941
ADC, adenocarcinoma; ALK, anaplastic lymphoma kinase; CDKN2A, cyclin-dependent kinase inhibitor 2A (which encodes INK4A and ARF); DDR2, discoidin
domain-containing receptor 2; Del19, EGFR exon 19 deletion; DNMT3A, DNA (cytosine‑5-)-methyltransferase 3α; EGFR, epidermal growth factor receptor;
EZH2, enhancer of zeste homologue 2; FGFR1, fibroblast growth factor receptor 1; GEMM, genetically engineered mouse model; Ins20, EGFR exon 20 insertion;
KEAP1, kelch-like ECH-associated protein 1; LKB1, liver kinase B1; MLL2, mixed-lineage leukaemia 2; N/A, not available; NF1, neurofibromin 1; NRG1, neuregulin 1;
NTRK1, neurotrophic tyrosine kinase, receptor, type 1; PDGFRA, platelet-derived growth factor receptor-α; PIK3CA, PI3K catalytic subunit-α; SCC, squamous cell
carcinoma; SOX2, SRY-box 2; TET2, TET methylcytosine dioxygenase 2. *Status refers to mechanisms by which each gene is altered in tumours — mutation (M),
copy number gain (C) or fusion (F).

NATURE REVIEWS | CANCER VOLUME 14 | AUGUST 2014 | 537

© 2014 Macmillan Publishers Limited. All rights reserved

REVIEWS
Endothelial PDGF Macrophage
cell
VEGF ECM: keratin,
Fibroblast fibronectin
and collagen
Erythrocyte
Tumour cell
PDL1
Neutrophil PD1
CD8+
T cell
CXCR2 CXCL
Figure 1 | The lung cancer microenvironment.  The tumour microenvironment,
Nature Reviews | Cancer
including endothelial cells, fibroblasts and myeloid cells, among others, has important
roles in determining the characteristics of lung tumours. It is likely that a combination
of the cell of origin, genetic alterations and microenvironmental factors all contribute to
the lineage identity of lung tumours. Extracellular matrix (ECM), which often consists of
keratins in lung squamous cell carcinoma and fibronectin in desmoplastic lung
adenocarcinomas, gives structural support to tumour cells and is associated with
tumour-associated fibroblasts. Blood vessels are newly formed at the tumour site by
recruitment of endothelial cells via platelet-derived growth factor (PDGF) and vascular
endothelial growth factor (VEGF), among others. As the blood and lymphatic vessels
form, numerous blood cells, including macrophages, neutrophils, T cells and B cells,
home to the tumours. In particular, tumours can recruit neutrophils through secretion of
CXC-chemokine ligand (CXCL) family members, which bind to the neutrophil receptor
CXCR2. In addition, tumour cells often express immune checkpoint molecules, such as
programmed cell death 1 ligand 1 (PDL1), to attenuate a cytotoxic response from T cells.
PD1, programmed cell death 1.
Tumours can evade immune surveillance by
expressing molecules that maintain tolerance to nor-
mal peripheral tissues, including the interaction of the
tumour-associated programmed cell death 1 ligand 1
(PDL1) with the immune receptor programmed cell
death 1 (PD1; also known as PDCD1). Recently, the
use of antibodies targeting the PD1–PDL1 checkpoint
has resulted in some marked responses in early-stage
clinical trials for a large panel of therapy-refractory can-
cer subtypes, including advanced melanoma, NSCLC
and renal cell cancer, with a proportion of responding
patients showing persistent long-term benefits44,45. The
PD1–PDL1 interaction inhibits CD8+ cytotoxic T lym-
phocyte proliferation, survival and effector function,
and can induce apoptosis of tumour-infiltrating T cells;
PD1–PDL1 interactions can also promote the differen-
tiation of CD4+ T cells into forkhead box P3‑expressing
(FOXP3+) regulatory T (TReg) cells, which are known to
further suppress the immune system and cause periph-
eral immune tolerance in patients with lung cancer 46.
Despite the promising clinical benefits, there is currently
no defined subset of patients with lung cancer who are
particularly sensitive to PD1–PDL1 blockade. This is
partly due to a lack in the understanding of how tumour
cells affect their microenvironment, including the sur-
rounding immune cells44,45,47. Evaluating the expres-
sion of PDL1 on tumour cells is only the starting point
in the analysis of the interactions between tumour cells
and the surrounding microenvironment 48,49. Many
important questions remain, including whether lung
tumours with distinct genetic backgrounds differ in how
they shape their immune microenvironment.
538 | AUGUST 2014 | VOLUME 14 www.nature.com/reviews/cancer
© 2014 Macmillan Publishers Limited. All rights reserved
Differences in the ability to secrete inflammatory
cytokines such as interleukin‑6 (IL‑6) may be one way
in which tumour cells influence their surroundings50,51.
Tumours that are driven by different oncogenic muta-
tions in mice, such as EGFR and Kras, have distinguish-
able immune infiltrates with respect to cell types and
their phenotypes in the tumour immune microenviron-
ment 48,49. These mechanisms have not been defined in
detail, and there are other important questions to con-
sider: does each genetic subset of NSCLC have its own
unique microenvironmental influences, or can common
characteristics of how the surroundings drive tumour
subsets be uncovered? How does targeted therapy alter
the tumour microenvironment? Do drug-resistant or
recurrent tumours have an environmental milieu that is
distinct from the initial untreated tumour? A more thor-
ough understanding of these dynamic interactions will
help to show new targets that can be manipulated to pro-
mote antitumor effects. Importantly, many of these ques-
tions are challenging to understand, given the need to
study the immune system in vivo, and the use of mouse
models with intact immune systems in combination with
patient samples may be instructive.
Cell(s) of origin for NSCLC heterogeneity at tumour ini-
tiation. Another contributing factor to the diversity of
NSCLCs may be the potential distinct cells of origin in
which subsets of NSCLC first arise. The cell of origin for
each subset of NSCLC is essentially unknown beyond
initial work in this area in mouse models. For example,
it remains to be understood whether multiple cell types
are equally likely to produce KRAS-mutant ADC or if
only one cell type exists in the right microenvironment
and must gain oncogenic KRAS expression to produce
this type of ADC. It is possible that the biology of differ-
ent cells of origin is what drives the different phenotypes
of NSCLCs with distinct genotypes. This could be the
result of unique gene expression patterns of the originat-
ing cells, differences in the type of cells that the originating
cells can produce, or unique microenvironments of the
originating cell type. Ultimately, the clinical importance
behind these seemingly basic biological questions is
whether a different cell of origin partly dictates treat-
ment responses. Can knowledge of the cell of origin pre-
dict environmental influences that can be targeted for
antitumour therapy? Furthermore, can knowledge of the
cell of origin be used for the earlier detection of tumours?
The answers to these questions have the capacity to
revolutionize our current concept of the stratification,
diagnosis and treatment of NSCLC.
A long-standing hypothesis proposes that stem and
progenitor cells in adult tissues function as carcinoma
cells of origin because they are the only cells that have a
sufficient lifespan to accumulate the many genetic altera-
tions required for tumour progression52. Furthermore,
stem cells have inherent self-renewal capacity and may
not need extensive epigenetic reprogramming. However,
even genetically normal cells with limited self-renewal
capacity can be induced to acquire more stem cell-like
properties in response to genetic alterations or micro­
environmental changes53,54, and this supports the idea that

Box 1 | Mouse models
Genetically engineered mouse models (GEMMs) have enabled numerous studies of
non-small-cell lung cancer (NSCLC) that would not be possible using patient samples or
cancer cell lines: for example, preclinical or co‑clinical trials of targeted therapies, the
study of metastatic and transplanted disease and examination of tumour cells of
origin25,49,74,98,125,126. Today, GEMMs for most of the common NSCLC driver mutations
have been generated, including for KRAS, epidermal growth factor receptor (EGFR),
and echinoderm microtubule-associated protein-like 4 (EML4)–anaplastic lymphoma
kinase (ALK); and despite their lack of genetic complexity compared to human cancers,
they have shown some remarkable similarities in pathological features and treatment
responses to the human disease98,125–127.
GEMMs are particularly informative when the net effects of several mutations need to
be determined in vivo. For example, the conditional oncogenic KrasG12D mouse model
has been used to elucidate the steps from early to late tumorigenesis, owing to the
temporal control it affords128, and it is easy to combine with mice bearing conditional
null alleles for other genes of interest. For example, KrasG12D tumours only reach a full
adenocarcinoma stage with a very long latency, but KrasG12D-expressing and Trp53‑null
tumours are more advanced and show a decreased response to certain treatment
strategies when compared to KrasG12D tumours128,129. Simultaneous inactivation of
Pten and liver kinase B1 (Lkb1) in the lung produced only squamous cell carcinoma
(SCC)49, and this fits with the preclinical observations that PI3K and mTOR pathways are
activated in most human lung SCC tumours20,49,65. Similar genetic breeding schemes can
be used to identify and validate potential treatment targets through in vivo synthetic
lethal experiments. Elegant studies have recently shown that MYC, cyclin-dependent
kinase 4 (CDK4) and CRAF are crucial KRAS effectors that can lead to synthetic lethality
when genetically inactivated in tumours with activated KRAS130–134.
The assessment of immunotherapeutics and the dynamic interactions between
tumour cells and their microenvironment using GEMMs (which are immunocompetent)
is another emerging research direction. Experiments of particular interest include gene
expression and pathway activation profiles for each cell type within the tumour;
genotype- or treatment-dependent influences on the tumour microenvironment; and
effects of individual or combination therapies on tumour cells, immune cells and other
cell types within the tumour microenvironment.
Patient-derived xenograft (PDX) models provide an alternative and complementary
method to GEMMs to address human–murine differences and allow for expansion of
patient material to perform assays such as metabolomic and serial transplantation135.
A ‘humanized’ lung and even a ‘humanized’ immune system in the mouse might offer a
more accurate means to model NSCLC.
more mature, differentiated cells may be just as likely to
give rise to malignancy. Historically, ADCs have been pro-
posed to arise from club cells (previously known as Clara
cells) or alveolar epithelial type 2 (AT2) cells, owing to
the staining of patient ADCs by immunohisto­chemistry
with markers of these cell types4,5. However, it is impor-
tant to note that the staining pattern of a tumour is merely
a snapshot of the gene expression of the tumour cells
at that time point and might not match the initiating
cell type. Our current understanding of cells of origin
for lung cancer is mostly derived from experimental
data using GEMMs55 (BOX 1; FIG. 2). Many conditional
GEMMs target activation and/or loss of genes specifically
to lung cells by intranasal or intratracheal instillation of
adenovirus-Cre, which infects lung epithelial cells along
the proximal to distal tract. After using intra­nasal adeno­
virus-Cre to induce oncogenic Kras, loss of Pten or loss
of p38, the first hyperproliferative cells to be observed
were bronchioalveolar stem cells (BASCs) — implicat-
ing them as possible ADC cells of origin56–58. However,
in more recent studies that targeted the expression of
oncogenic KrasG12D only in cells expressing club cell
secretory protein (CCSP), such as club cells and BASCs,
or only in cells expressing surfactant protein C (SPC),
NATURE REVIEWS | CANCER VOLUME 14 | AUGUST 2014 | 539
© 2014 Macmillan Publishers Limited. All rights reserved
REVIEWS
such as AT2 cells and BASCs, AT2 cells seemed to be the
only cells that were capable of giving rise to advanced
ADC in the alveolar space, whereas club cells and BASCs
seemed to be limited to driving bronchiolar hyperplasia
within the same time frame59. Using these approaches, it
is notable that changes in the cells of origin were evident
when the genotype for tumour initiation was altered (for
example, to include p53 loss) or if injury or inflamma-
tion were present during tumour initiation (for example,
after adenovirus infection or after naphthalene-induced
injury)56,59–63. Injury or inflammation probably more
closely mimics the scenario of tumour initiation in
humans, in which environmental influences and ongo-
ing injury occur in contrast to the relatively sterile mouse
colony. These questions are unexplored in other models
of ADC that use distinct oncogenes or in SCC. Thus, it
remains entirely possible that club cells, AT2 cells and
BASCs are all possible initiators of lung ADC. Future
development of more precise lineage-specific Cre driv-
ers combined with approaches to study tumorigenesis
in the context of injury and inflammation (situations
that are more likely to mimic carcinogenesis in humans)
will be needed to better determine the comparable
ADC-initiating activity of these populations.
Although increasing amounts of genomic data show
that distinct gene expression programmes and driver
mutations distinguish ADC from SCC, it remains unclear
whether these two tumour types arise from a common
cell of origin or diverse cell types, including different lung
stem or progenitor cells (BOX 2). Until recently, a paucity of
GEMMs for SCC has precluded analysis of the cells
of origin of this important NSCLC subtype. It has long
been hypothesized that SCC arises from basal cells, as
lung SCCs most frequently arise in the proximal lung 4,
but also because they often express KRT5, SOX2 and
p63, which are markers of the normal basal cell popula-
tion5–7,62,64. GEMMs of ADC have been more widespread,
mostly owing to the usefulness and availability of the con-
ditional oncogenic Kras allele (which drives lung ADCs
both independently and more rapidly in combination
with Trp53 loss) as well as early models using chemicals
that induce RAS mutations to drive tumours. Although
KRAS or NRAS mutations are present in up to 25% of
ADCs, they are rarely detected in SCCs, and mouse
modelling with these oncogenes seems to result predomi-
nantly in the development of ADC. Mutations that are
common in samples from patients with SCC have only
recently been catalogued, and this opens up new ideas
about how to model SCC20. Kinase-dead IKKα knock‑in
mice develop spontaneous lung SCC, but because this
mouse has a germline Ikka mutation, it is not clear which
cells in the lung expanded into the squamous tumours38.
Loss of the tumour suppressor liver kinase B1 (Lkb1; also
known as Stk11) in the oncogenic KrasG12D model pro-
duces a mixture of tumours, including ADC, SCC and
large cell carcinoma65. Similarly, a mixture of ADC
and SCC is found in mice after targeted deletion of Pten
or transforming growth factor-β receptor 2 (Tgfbr2)
in proximal cells with keratin-driven Cre alleles in the
KrasG12D background66. Expression of the transcription
factor Sox2 (overexpressed in 20–60% of human SCCs) in
REVIEWS

Proximal Basement Pseudostratified

membrane columnar epithelium
Cartilage Club cell (CCSP+)
Ciliated cell (AcTUB+)
Goblet cell
Basal cell (KRT5+ and p63+)

Typical SCC Modelled by:

• SOX2+ • KrasG12D expression
• p63+ and Lkb1-null
• KRT5+ • Lkb1-null and Pten-null
• IKKα: kinase dead

Alveolar space AT2 (SPC+)

BASC
(CCSP+ and SPC+)
AT1
Distal

Simple columnar epithelium

Typical ADC Modelled by:
• KrasG12D
Neuroendocrine • TTF1+ • KrasG12D expression
cell (CGRP+) • KRT7+ and Trp53-null
• SPC+ • EGFRT790M/L858R
• Many others

Figure 2 | A diagram of proximal and distal lung cells, indicating markers that are retained in carcinomas and
putative squamous cell carcinoma (SCC) and adenocarcinoma (ADC) cells of origin.  DiverseNature Reviews | Cancer
lung stem or progenitor
cell populations are thought to have the ability to drive lung oncogenesis in different contexts. In the proximal lung, the
tracheal basal cell has been proposed to be the cell of origin for lung SCC. The evidence for this relationship includes
the expression of p63, SRY-box 2 (SOX2) and keratin 5 (KRT5) within the basal cells, squamous metaplasia of the basal cells
(common in smokers), and squamous cell carcinomas. Squamous tumours are modelled in mice by KrasG12D expression and
liver kinase B1 (Lkb1) knockout (20% of lesions are squamous), knocking in a germline dominant-negative kinase-dead
inhibitor of nuclear factor-κB kinase subunit-α (IKKα) and knocking out both Lkb1 and Pten (100% of lesions are squamous
for the second two models). Two bronchiolar cell populations, the bronchiolar progenitor cells and the bronchioalveolar
stem cells (BASCs) may also be able to give rise to tumours with squamous characteristics, although experimental lineage
tracing is needed to confirm this theory. ADCs can be modelled by KrasG12D expression (long latency), KrasG12D expression
and Trp53‑null, and epidermal growth factor receptor (EGFR)T790M/L858R, among other genetic models, and they are thought
to arise from more proximal airway cells. These tumours often retain characteristics of proximal airways, such as the
expression of surfactant protein C (SPC), KRT7 and thyroid transcription factor 1 (TTF1). Again, BASCs or bronchiolar
progenitor cells, which are able to give rise to alveolar lineages after lung injury, may likewise be able to give rise to
tumours with alveolar characteristics. AcTUB, acetylated tubulin; AT, alveolar epithelial type; CCSP, club cell secretory
protein; CGRP, calcitonin gene-related peptide.

club cells and BASCs produces lung tumours that express
the marker p63 but histologically resemble ADCs7. This
intriguing finding suggests that distal lung epithelia are
unable to produce a fully squamous phenotype, despite the
expression of an SCC transcription factor. In addition,
the deletion of Lkb1 and Pten in the lung via intranasal
adenovirus-Cre was recently shown to give rise to fully
penetrant lung SCCs49. The next important step will be to
use lineage-restricted Cre alleles, such as the oestrogen-
responsive Cre under control of the Krt5 promoter
(Krt5–Cre-ER), to determine which lung cells that are
null for Lkb1 and Pten are able to drive squamous disease.
Tumour-propagating cells (TPCs) and cellular plastic-
ity: heterogeneity between tumour cells. ‘Cancer stem
cells’ or TPCs, which are defined as the tumour cells
with the stem cell-properties of self-renewal and dif-
ferentiation, have the capacity to produce tumours in
transplantation assays. Establishment of tumours at
metastatic sites and tumour recurrence following treat-
ment have been attributed to growth and survival of
TPCs67,68. Recent studies have identified potential cell
surface markers or genetic traits that may mark the TPC
population in NSCLC, such as aldehyde dehydroge-
nase (ALDH) activity or expression of NOTCH, CD24,
CD166 or CD44 (REFS 38,69–73). However, these stud-
ies have not used serial transplant assays in the context
of the lung environment, and a bona fide human lung
TPC remains to be defined. GEMMs have allowed for
more systematic study of lung TPC phenotypes, includ-
ing serially transplanted tumours and metastases. Studies
in the KrasG12D-expressing and Trp53‑null model of ADC
suggest that stem cell antigen 1 (SCA1, also known as
Ly6A)+, CD24+, β4 integrin+, and NOTCH3hi mark
the TPC population70,73. The identity of TPCs from
other ADC GEMMs is unknown; SCA1 did not enrich

540 | AUGUST 2014 | VOLUME 14 www.nature.com/reviews/cancer

© 2014 Macmillan Publishers Limited. All rights reserved


Box 2 | Lung stem and progenitor cell populations
Genetic lineage tracing and cell biology approaches have shown that the murine lung
contains region-specific stem and progenitor cell populations that respond to local
injury. Basal cells function as stem cells for the trachea, main bronchi and upper airways,
where they can replace the pseudostratified epithelium, including secretory club cells
(previously known as Clara cells), mucus-producing goblet cells and ciliated cells136–138.
In more distal airways, club cells are a self-renewing population that maintains the
ciliated cells139, and subsets of club cells, such as bronchiolar progenitors, can give rise
to ciliated and club cell lineages after injury140,141. In the alveolar space, where gas
exchange is carried out by alveolar epithelial type 1 (AT1) cells, the surfactant-
expressing AT2 cells can function as stem cells60,142. Another alveolar cell
population, expressing α6β4 integrin, can also produce alveolar epithelia142,143.
Bronchioalveolar stem cells (BASCs), which reside between the airway and alveolar
space, can give rise to both epithelial lineages56,114,144–146. Murine proximal and distal
lung stem cells can be isolated by fluorescence-activated cell sorting that uses different
cell surface markers and can be grown in three-dimensional culture systems to study
their differentiation potential114,136,141,147. Basal cells can be isolated from mouse or
human lung on the basis of their expression of nerve growth factor receptor
(NGFR)136,148, and AT2 cells can be purified from distal lung — most recently with the
marker HTII‑280 (REF. 142). Several other human lung stem cell populations have been
reported in the human lung, but their roles have been controversial149, and this points to
the characterization of human lung stem and progenitor cells as an important area for
future research. Furthermore, precisely how these cell types change their lineage
potential in the face of oncogenic insult coupled with injury is unknown and is likely to
influence tumorigenesis; injury and transformation might substantially alter plasticity53.
A better understanding of lung stem and progenitor cells and methods for their analysis
would open up new ways to explore the cellular origins of lung tumorigenesis.
for TPCs from the Kras- or EGFR-driven GEMMs74.
In the first lung-specific genetic model of SCC (the Lkb1-
and Pten-null model) the TPCs had a high expression
of SCA1 and the basal cell marker nerve growth factor
receptor (NGFR). Intriguingly, these TPCs also expressed
high levels of the immune-checkpoint molecule PDL1,
which may be targetable as described above49. Overall,
these findings indicate the importance of taking the
geno­type of the tumour into account when seeking to
define a TPC population; each subset of NSCLC might
harbour TPCs with unique surface markers and molec-
ular drivers, which could each be uniquely targeted.
Alternatively, many subsets of NSCLC might not have
one TPC population that can be defined. Future research
focusing on resolving the metastatic activity and therapy
response of murine TPCs and the molecules that control
them may help to translate these findings to improve the
treatment of patients with lung cancer.
The genetic complexity and rapid clonal evolution
of lung tumours could mean that if TPCs do occur in
most lung cancers, they will have a high degree of plas-
ticity. Fascinating clinical observations have shown some
patients who are initially diagnosed with EGFR-driven
Pseudostratified epithelium ADCs develop SCLC after long-term treatment with the
This describes the epithelium EGFR tyrosine kinase inhibitors gefitinib or erlotinib75,76.
of the trachea, which is truly a In contrast to ADC models, lineage tracing and viruses
monolayer but appears to have
that are engineered to express Cre under the control of
some stratification due to the
variable distances of the nuclei various cell-type-specific promoters have been used to
from the basal lamina. show that SCLCs probably arise from neuroendocrine
cells76–78. However, examination of these tumours before
Patient-derived xenograft and after SCLC conversion shows the persistence of the
(PDX). Primary tumour cells
from fresh patient tumours that
same EGFR mutations, suggesting a shared clonal ori-
are propagated subcutaneously gin of both types of tumours. This highlights the poten-
in immunocompromised mice. tial epigenetic plasticity of lung cells and lung tumours
NATURE REVIEWS | CANCER VOLUME 14 | AUGUST 2014 | 541
© 2014 Macmillan Publishers Limited. All rights reserved
REVIEWS
after therapy 75. Further careful evaluation of TPC activ-
ity and cellular plasticity of tumour cells with patient
tissues, probably using patient-derived xenograft (PDX)
models and GEMMs of lung cancer, will help us to better
understand tumour lineage conversion as a path towards
developing chronic treatment resistance. These findings
also highlight the importance of considering how cells of
origin may differ, depending on the therapeutic status
of the tumour environment.
Integrated therapies for NSCLC
Target validation and patient stratification. Although
studies of lung cancer genomes have implicated several
genes as likely crucial mediators of tumour initiation and
progression, experimental validation of the most impor-
tant, functional genomic changes in lung cancer cells
remains a challenge. Despite computational approaches
to separate ‘driver’ alterations from passenger altera-
tions, this distinction is probably more nuanced, and
substantial work will need to be completed to model the
consequences of specific genome alterations in NSCLC.
Existing repositories of lung cancer cell lines, as well as
efforts to generate new cell lines from patient tumours
have led to a number of important discoveries, although
these cell lines still fail to represent the full diversity of
human NSCLCs79. Three-dimensional culture techniques
might also offer a new way to propagate normal and
tumorigenic lung cells to better probe vulnerabilities of
tumour cells49,73. Multiple in vivo models using mice to
recapitulate lung cancer disease processes and treatment
responses have been generated, including GEMMs har-
bouring specific genetic aberrations that have been iden-
tified in human tumours55,80 (BOX 1). Translation of the
experimental results obtained through in vitro and in vivo
modelling systems has formed the basis for current and
future patient stratification paradigms (BOX 3). The limi-
tations of these approaches should also be considered in
future work to develop a more precise understanding of
how to predict therapy response.
Current treatments for NSCLC. The past decade has
seen some truly impressive new treatments for subsets of
patients with lung cancer, most of whom harbour muta-
tions in one of the key oncogenic driver mutants upon
which tumour survival and progression are depend-
ent. These include mutations in EGFR, the echinoderm
microtubule-associated protein-like 4 (EML4)–ALK fusion
and ROS1 fusions81,82. Extensive preclinical and clinical
studies have proven the marked treatment responses and
survival advantages over conventional chemotherapies
that are provided by target-specific inhibitors to EGFR-
activating mutations or to ALK fusions 83–85. Recent
genomic analyses of lung SCC have also given the first
set of potentially targetable driver mutations, including
FGFR1, FGFR2, FGFR3, DDR2 and PI3K20. Clinical trials
that aim to target these subsets of patients who have Stage
I–IIIA lung cancers are currently underway; preliminary
results were presented at the 2014 American Association
for Cancer Research Annual Meeting86, and these showed
responses to an FGFR inhibitor (BGJ398) in a subset of
patients with SCC who have FGFR1 amplification.
REVIEWS
Box 3 | Patient stratification
Stratification and treatment selection for patients with non-small-cell lung cancer
(NSCLC) heavily relies on radiographical and pathological evaluation in standard
clinical practice. In recent years, molecular diagnostic platforms have been gradually
introduced into this process. Today, many cancer centres and hospitals have adopted
some degree of genetic diagnosis. Commonly accepted oncogenic driver mutations,
including KRAS, epidermal growth factor receptor (EGFR), anaplastic lymphoma kinase
(ALK), ROS and BRAF, are being sequenced and detected as a standard diagnosis
procedure. Increasingly, mutation-based decision-making procedures are being
integrated in the clinic, and we expect that additional novel technology platforms that
stratify tumours according to the specific metabolome, epigenome and immune profile
of each patient will be applied in the clinic in the near future. The anticipated challenge
is how best to verify and use the vast amounts of sequencing information for translation
to the clinic. Ongoing efforts seek to optimize data mining that will link existing
genomic and biological data with clinical databases. In 2013, the Broad Institute,
Cambridge, Massachusetts, USA, launched a global alliance that intends to share
genomic and clinical data. A similar effort at Vanderbilt University, Nashville, Tennessee,
USA, which is mediated by a publicly accessible website (My Cancer Genome),
emphasizes the clinical application of cancer research. Worldwide efforts, such as the
International Cancer Genome Consortium (ICGC) and the Catalogue of Somatic
Mutations in Cancer (COSMIC) from the Sanger Institute, Hinxton, UK, and joint efforts
in European countries to establish organoid cultures from primary tumours or biopsies
from patients are also under way. Despite these independent efforts to integrate data
sets, a more organized programme is needed on the national and international levels.
The US National Center for Biotechnology Information recently initiated
whole-genome sequencing to identify rare, druggable oncogenic events in patients
who showed isolated but marked responses to certain drugs; this may represent the
first exploratory step towards an integrated programme.
Unfortunately, acquired resistance to chronic treat-
ment often develops within 9–12 months in most patients
who are treated with kinase inhibitors84,87,88. Therefore,
patients who have Stage I–IIIA tumours are still treated
by surgical resection as a first-line treatment and receive
combination chemotherapy as a standard of care, with
the use of targeted agents still considered to be experi-
mental. For patients with advanced disease who have
progressed on an inhibitor of EGFR or ALK, several
recurrent secondary mutations have been identified, such
as EGFR‑T790M and additional kinase domain mutations
in ALK87,88. Hence, finding treatments for tumours that
are resistant to first-generation EGFR or ALK inhibi-
tors has been a recent focus. Several newly developed
inhibitors that either have more potency or are ration-
ally designed to favourably target the mutated kinases,
such as AZD9291 and CO‑1686 for EGFR and LDK378
for EML4–ALK, have generated promising initial clini-
cal results89–91. Discovery of the mechanisms that under-
lie acquired resistance in patients without additional
mutations in the primary driver gene is also greatly
facilitated by high-throughput analytical approaches.
Amplifications of ALK and alternative pro-cancerous
pathway activations were identified in ALK fusion-
positive patients who have become resistant to the first-
generation ALK inhibitor crizotinib87. In patients who
are resistant to chronic EGFR inhibitor treatment, a wide
EGFR‑T790M range of resistance mechanisms has been reported. These
The most common mutation include increased activities of additional kinases owing
(~50%) in the epidermal to MET, HER2 or ERK amplification, additional muta-
growth factor receptor (EGFR)
gene that confers resistance to
tion of PIK3CA (which encodes the PI3K p110α sub­
EGFR tyrosine kinase inhibitors unit) or overexpression of AXL kinase14,92–95. Enhanced
such as erlotinib and gefitinib. NF‑κB signalling activity was also implied as one possible
542 | AUGUST 2014 | VOLUME 14 www.nature.com/reviews/cancer
© 2014 Macmillan Publishers Limited. All rights reserved
resistance mechanism that is evident by an improved
response and survival in patients with EGFR mutations
who have an increased expression of the NF‑κB inhibi-
tor IκBα (also known as NFKBIA)96. In addition, a com-
mon BIM (also known as BCL2L11) polymorphism that
results in changes in splicing and the deletion of the pro-
apoptotic BCL‑2‑homology domain (BH3) was shown to
potentially mediate intrinsic resistance to EGFR inhibi-
tors97, highlighting the complexity of possible resistance
mechanisms. It is conceivable that the comprehen-
sive acquisition of information on different aspects of
tumour biology, such as genomic and kinase profiling in
patients, will be crucial in the future to determine the best
course of treatment following any new diagnosis or the
development of acquired resistance.
Most patients with advanced stage NSCLC without
targetable genomic alterations are still treated by conven-
tional chemotherapies. Activating KRAS mutations were
identified and verified long before the discovery of mutant
EGFR. However, treatment choices for patients with
KRAS-mutant lung cancer are still very limited. Current
efforts to treat this subset of patients have been mostly
focused on inhibiting common KRAS downstream sig-
nalling cascades. The RAF–MEK–ERK pathway, which
is activated directly downstream of KRAS, has proven to
be a valid target in both preclinical models and clinical tri-
als98–100. However, the clinical benefits of MEK inhibitors,
even in combination with other agents, are still somewhat
moderate compared to those of target-specific inhibi-
tors such as erlotinib for patients with activating EGFR
mutations, and the use of MEK inhibitors is associated
with additional complications and enhanced toxicity 100.
The available preclinical and clinical results present clear
challenges to the common belief that therapies target-
ing one or a few specific alterations should have fewer
side effects and lower toxicity compared to standard
chemotherapies. Indeed, this is not entirely a surprise,
as many of the targeted pathways for lung cancer treat-
ment are also essential for normal tissue functions. The
simultaneous inhibition of multiple signalling pathways
can be deleterious to necessary normal cells. One pos-
sible remedy being explored is to optimize treatment
schedules and improve targeting efficiencies for single-
pathway inhibition by improving inhibitor potency or
linear inhibition of multiple targets within the pathway.
Nonetheless, alternative treatment approaches with less
toxicity and better responses are of immediate need. A few
studies have more recently reported the rational design
of KRAS inhibitors that target the cysteine residue of the
common KRAS mutation G12C in lung cancer 101,102, and
these are therefore similar to the second-generation EGFR
inhibitors (such as WZ4002, AZD9291 and CO‑1686)
that target EGFR‑T790M. In vitro studies of these KRAS
inhibitors demonstrate a proof of concept 101,102; however,
the in vivo efficacy of these molecules still requires much
more investigation.
Targeting multiple cellular compartments in lung can-
cer. Similar to KRAS mutations, many newly identified
potential pro-cancerous changes, such as overexpression
of the transcription factors SOX2 and MYC103–105, present
 REVIEWS

Cytotoxic T lymphocyte
protein 4
(CTLA4; also known as
CD152). A surface receptor
that transmits inhibitory
signals to T cells.
CD73
A cell surface enzyme that
generates extracellular
adenosine, which inhibits
T cell function.
CD47
The receptor for
thrombospondin 1 (TSP1).
CD47 is highly expressed in
many tumour cells.
Chimeric antigen receptors
(CARs). Genetically engineered
receptors that result in desired
specificity (to tumour cells) in
effector T cells.
CpG island methylator
phenotype
(CIMP). Reflects the genomic
status that multiple CpG
islands are methylated
simultaneously, leading to
epigenetic inactivation of
different genes, including
tumour suppressors.
clear challenges to our current ideas about treatment
approaches — in cases in which there is no clear drugga-
ble target, what can be done? Furthermore, the short-lived
in vivo efficacy for most if not all existing small molecule
inhibitors87,106 also advocates more durable treatment
approaches. On the basis of our current understanding,
the more effective approach probably requires therapies
that not only target tumour cells but also target other
components of the tumour, such as tumour vasculature,
tumour-associated fibroblasts and tumour-specific and/or
non-specific immune cells. Besides the more recently
studied PD1–PDL1 inhibitory pathway, other approaches
that intervene with the immune system, such as antibod-
ies against cytotoxic T lymphocyte protein 4 (CTLA4; also
known as CD152), CD73 or CD47, and more sophisticated
cellular immune therapies, such as engineered T cell
therapy using chimeric antigen receptors (CARs), are also
under extensive scrutiny 107–110. More importantly, ongo-
ing efforts are seeking to discover the best combination
approach that integrates immune therapy with other
therapies. Angiogenesis has long been seen as a possible
therapeutic window, with many novel therapeutic agents
that have been developed or are being developed to clini-
cally target this process, although the overall efficacy of
anti-angiogenic agents has been modest in unselected
patient cohorts111. Emerging evidence has suggested that
the combination of immunotherapy and anti-angiogenic
agents has potential synergistic effects112,113, pointing to
a new possible avenue to mutually enhance both treat-
ments. In addition to providing key nutrients and oxy-
genated blood, tumour vasculature might have a role
in supporting TPCs35,114. Similarly, stromal cells such as
fibroblasts have been shown to provide additional signals
that support tumour growth and survival, and they may
therefore have major roles in primary and acquired treat-
ment resistance35,115. Understanding how best to target
these various aspects of the tumour microenvironment
would require a high-throughput comparison of changes
in the tumour microenvironment that occur upon single
and combination treatments.
Targeted therapies might also be able to indirectly
regulate tumour growth. Two prominent examples are
drugs that target epigenetic enzymes and metabolic
enzymes. Targeting epigenetic enzymes is expected to
enable marked perturbation of gene expression within
tumour cells to stop tumour growth. The recently devel-
oped bromodomain protein inhibitors have shown effi-
cacy in numerous preclinical studies116, including in lung
cancer 117, and they are currently under evaluation in the
clinic, including in the ongoing Phase I clinical trials
NCT01987362 and NCT01587703. Variations in expres-
sion as well as recurrent mutations were also reported
for several histone- and DNA-modifying enzymes,
including enhancer of zeste homologue 2 (EZH2), TET
methyl­cytosine dioxygenase 2 (TET2) and DNA methyl­
transferase 3A (DNMT3A) in all subtypes of NSCLC28.
In a similar concept, altered metabolism is one of the key
features of cancer cells. Anti-diabetic drugs, insulin-like
growth factor 1 receptor (IGF1R) inhibitors and drugs
that target glycolysis or lipid, nucleic acid and amino acid
synthesis are currently being explored for anti­tumour
activities in NSCLC118–121. Targeting metabolism is cer-
tainly promising for cancer control, particularly when
combined with other approaches. Recent studies have also
highlighted connections between TET and IDH, which
could have resulted in a CpG island methylator phenotype
(CIMP) in a subset of lung cancer, and this ‘connects the
dots’ between epigenetics and metabolism122–124.
Conclusion
The quickened pace of discovery of mechanisms that
underlie lung cancer development and possible treatments
in the past decade present the opportunity to integrate
information from multiple approaches for future lung
cancer treatment. Large amounts of information about the
identity of individual lung tumours are being collected.
New and improved functional studies are needed to meet
the pace of data set generation, and all of the aspects of
tumour heterogeneity — genetic, cellular and epigenetic
— need to be integrated to determine better approaches to
make an impact in this devastating disease. We anticipate
the future treatment scheme to be a genotype-depend-
ent, carefully selected combination that would ensure an
enhanced tumour immune reaction, inhibition of angio-
genesis and blockade of interactions between tumour cells
and stromal cells. Thus, we advocate ‘integrated therapy’,
in contrast to the current concept of targeted therapy, as
the future of effective NSCLC treatment.

1. Jemal, A. et al. Global cancer statistics. CA Cancer
J. Clin. 61, 69–90 (2011).
2. Siegel, R., Naishadham, D. & Jemal, A. Cancer
statistics, 2013. CA Cancer J. Clin. 63, 11–30 (2013).
3. Ettinger, D. S. et al. Non-Small Cell Lung Cancer,
Version 2.2013. J. Natl Compr. Canc. Netw. 11,
645–653 (2013).
4. Davidson, M. R., Gazdar, A. F. & Clarke, B. E. The
pivotal role of pathology in the management of lung
cancer. J. Thorac. Dis. 5, S463–S478 (2013).
5. Langer, C. J., Besse, B., Gualberto, A., Brambilla, E. &
Soria, J.‑C. The evolving role of histology in the
management of advanced non–small-cell lung cancer.
J. Clin. Oncol. 28, 5311–5320 (2010).
6. Rock, J. R., Randell, S. H. & Hogan, B. L. M. Airway
basal stem cells: a perspective on their roles in
epithelial homeostasis and remodeling. Dis. Model.
Mech. 3, 545–556 (2010).
7. Lu, Y. et al. Evidence that Sox2 overexpression is
oncogenic in the lung. PLoS ONE 5, http://dx.doi.
org/10.1371/journal.pone.0011022 (2010).
8. Davies, H. et al. Mutations of the BRAF gene in human
cancer. Nature 417, 949–954 (2002).
9. Santos, E. et al. Malignant activation of a K‑ras
oncogene in lung carcinoma but not in normal
tissue of the same patient. Science 223, 661–664
(1984).
10. Lynch, T. J. et al. Activating mutations in the epidermal
growth factor receptor underlying responsiveness of
non-small-cell lung cancer to gefitinib. N. Engl. J. Med.
350, 2129–2139 (2004).
11. Paez, J. G. et al. EGFR mutations in lung cancer:
correlation with clinical response to gefitinib therapy.
Science 304, 1497–1500 (2004).
In this study, the clinical observation that only
some lung cancers responded well to the EGFR
inhibitor gefitinib was linked to the presence of
EGFR-activating mutations in the responding
tumours. This finding revolutionized the
stratification of patients that receive EGFR
inhibitor treatment and furthered our
understanding of oncogene addiction in NSCLC.
12. Pao, W. et al. EGF receptor gene mutations are
common in lung cancers from “never smokers” and are
associated with sensitivity of tumors to gefitinib and
erlotinib. Proc. Natl Acad. Sci. USA 101,
13306–13311 (2004).
13. Shepherd, F. A. et al. Erlotinib in previously treated
non-small-cell lung cancer. N. Engl. J. Med. 353,
123–132 (2005).
14. Engelman, J. A. et al. MET amplification leads to
gefitinib resistance in lung cancer by activating ERBB3
signaling. Science 316, 1039–1043 (2007).
15. Fernandez-Cuesta, L. et al. CD74‑NRG1 fusions in lung
adenocarcinoma. Cancer Discov. 4, 415–422 (2014).
16. Kohno, T. et al. KIF5B‑RET fusions in lung
adenocarcinoma. Nature Med. 18, 375–377 (2012).
17. Rikova, K. et al. Global survey of phosphotyrosine
signaling identifies oncogenic kinases in lung cancer.
Cell 131, 1190–1203 (2007).
18. Soda, M. et al. Identification of the transforming
EML4‑ALK fusion gene in non-small-cell lung cancer.
Nature 448, 561–566 (2007).

NATURE REVIEWS | CANCER VOLUME 14 | AUGUST 2014 | 543

© 2014 Macmillan Publishers Limited. All rights reserved

REVIEWS
19. Stephens, P. et al. Lung cancer: intragenic ERBB2
kinase mutations in tumours. Nature 431, 525–526
(2004).
20. The Cancer Genome Atlas Research Network (TCGA).
Comprehensive genomic characterization of squamous
cell lung cancers. Nature 489, 519–525 (2012).
As part of The Cancer Genome Atlas, this study
was the first to provide a comprehensive landscape
of genomic and epigenomic alterations in lung
SCCs. This study identified potential druggable
targets in most tumours, and many of these targets
are now being validated as therapeutic options.
21. Vaishnavi, A. et al. Oncogenic and drug-sensitive
NTRK1 rearrangements in lung cancer. Nature Med.
19, 1469–1472 (2013).
22. Weiss, J. et al. Frequent and focal FGFR1 amplification
associates with therapeutically tractable FGFR1
dependency in squamous cell lung cancer. Sci Transl.
Med. 2, 62ra93 (2010).
23. Ding, L. et al. Somatic mutations affect key pathways
in lung adenocarcinoma. Nature 455, 1069–1075
(2008).
24. Imielinski, M. et al. Mapping the hallmarks of lung
adenocarcinoma with massively parallel sequencing.
Cell 150, 1107–1120 (2012).
25. Heist, R. S. & Engelman, J. A. Snapshot: non-small
cell lung cancer. Cancer Cell 21, 448–448.e2 (2012).
26. Guagnano, V. et al. FGFR genetic alterations predict
for sensitivity to NVP‑BGJ398, a selective pan-FGFR
inhibitor. Cancer Discov. 2, 1118–1133 (2012).
27. Hammerman, P. S. et al. Mutations in the DDR2
kinase gene identify a novel therapeutic target in
squamous cell lung cancer. Cancer Discov. 1, 78–89
(2011).
28. Kandoth, C. et al. Mutational landscape and
significance across 12 major cancer types. Nature
502, 333–339 (2013).
29. The Cancer Genome Atlas Research Network (TCGA).
Comprehensive molecular profiling of lung
adenocarcinoma. Nature http://dx.doi.org/10.1038/
nature13385 (2014).
30. Lawrence, M. S. et al. Discovery and saturation
analysis of cancer genes across 21 tumour types.
Nature 505, 495–501 (2014).
In this study, the genetic complexity of 21 tumour
types was directly compared, demonstrating that
both lung ADCs and lung SCCs have extremely high
somatic mutation rates compared with other
tumour types. These authors also present a
comprehensive catalogue of cancer genes for each
tumour subtype, furthering our global
understanding of cancer.
31. Govindan, R. et al. Genomic landscape of non-small
cell lung cancer in smokers and never-smokers. Cell
150, 1121–1134 (2012).
32. Barbie, D. A. et al. Systematic RNA interference
reveals that oncogenic KRAS-driven cancers require
TBK1. Nature 462, 108–112 (2009).
33. Sun, N. et al. Isocitrate dehydrogenase 1 is a novel
plasma biomarker for the diagnosis of non-small cell
lung cancer. Clin. Cancer Res. 19, 5136–5145
(2013).
34. Tan, F. et al. Identification of isocitrate
dehydrogenase 1 as a potential diagnostic and
prognostic biomarker for non-small cell lung cancer
by proteomic analysis. Mol. Cell Proteomics 11,
M111.008821 (2012).
35. Quail, D. F. & Joyce, J. A. Microenvironmental
regulation of tumor progression and metastasis.
Nature Med. 19, 1423–1437 (2013).
36. DuPage, M. et al. Endogenous T cell responses to
antigens expressed in lung adenocarcinomas delay
malignant tumor progression. Cancer Cell 19, 72–85
(2011).
By engineering KRAS-driven GEMM tumours to
express neoantigens, the authors of this study
explored the dynamic responses of endogenous
T cells to naturally arising tumours and informed
our ideas about how lung tumours interact with
immune cells.
37. Murdoch, C., Muthana, M., Coffelt, S. B. & Lewis, C. E.
The role of myeloid cells in the promotion of tumour
angiogenesis. Nature Rev. Cancer 8, 618–631
(2008).
38. Xiao, Z. et al. The pivotal role of IKKα in the
development of spontaneous lung squamous cell
carcinomas. Cancer Cell 23, 527–540 (2013).
39. Vignaud, J.‑M. et al. The role of platelet-derived
growth factor production by tumor-associated
macrophages in tumor stroma formation in lung
cancer. Cancer Res. 54, 5455–5463 (1994).
544 | AUGUST 2014 | VOLUME 14 www.nature.com/reviews/cancer
40. Zaynagetdinov, R. et al. A critical role for macrophages
in promotion of urethane-induced lung carcinogenesis.
J. Immunol. 187, 5703–5711 (2011).
41. Bellocq, A. et al. Neutrophil alveolitis in
bronchioloalveolar carcinoma: induction by tumor-
derived interleukin‑8 and relation to clinical outcome.
Am. J. Pathol. 152, 83–92 (1998).
42. Houghton, A. M. et al. Neutrophil elastase-mediated
degradation of IRS‑1 accelerates lung tumor growth.
Nature Med. 16, 219–223 (2010).
43. Mantovani, A., Cassatella, M. A., Costantini, C. &
Jaillon, S. Neutrophils in the activation and regulation
of innate and adaptive immunity. Nature Rev.
Immunol. 11, 519–531 (2011).
44. Brahmer, J. R. et al. Safety and activity of anti‑PD‑L1
antibody in patients with advanced cancer. N. Engl.
J. Med. 366, 2455–2465 (2012).
45. Topalian, S. L. et al. Safety, activity, and immune
correlates of anti‑PD‑1 antibody in cancer. N. Engl.
J. Med. 366, 2443–2454 (2012).
References 44 and 45 were the first studies to
show the efficacy and durable response of
immunotherapy for the treatment of late-stage
NSCLCs. The data suggest a relationship between
PDL1 expression on tumour cells and objective
response, leading to many new studies examining
the expression of this immune-evasion molecule
on different NSCLCs.
46. Keir, M. E., Butte, M. J., Freeman, G. J. &
Sharpe, A. H. PD‑1 and its ligands in tolerance and
immunity. Annu. Rev. Immunol. 26, 677–704 (2008).
47. Forde, P. M., Kelly, R. J. & Brahmer, J. R. New
strategies in lung cancer: translating immunotherapy
into clinical practice. Clin. Cancer Res. 20,
1067–1073 (2014).
48. Akbay, E. A. et al. Activation of the PD‑1 pathway
contributes to immune escape in EGFR-driven lung
tumors. Cancer Discov. 3, 1355–1363 (2013).
49. Xu, C. et al. Loss of Lkb1 and Pten leads to lung
squamous cell carcinoma with elevated PD‑L1
expression. Cancer Cell 25, 590–604 (2014).
50. Ancrile, B., Lim, K. H. & Counter, C. M. Oncogenic Ras-
induced secretion of IL6 is required for tumorigenesis.
Genes Dev. 21, 1714–1719 (2007).
51. Gao, S. P. et al. Mutations in the EGFR kinase domain
mediate STAT3 activation via IL‑6 production in
human lung adenocarcinomas. J. Clin. Invest. 117,
3846–3856 (2007).
52. Visvader, J. E. Cells of origin in cancer. Nature 469,
314–322 (2011).
53. Tata, P. R. et al. Dedifferentiation of committed
epithelial cells into stem cells in vivo. Nature 503,
218–223 (2013).
54. Chaffer, C. L. et al. Normal and neoplastic nonstem
cells can spontaneously convert to a stem-like state.
Proc. Natl Acad. Sci. 108, 7950–7955 (2011).
55. Kwon, M. & Berns, A. Mouse models for lung cancer.
Mol. Oncol. 7, 165–177 (2013).
56. Kim, C. et al. Identification of bronchioalveolar stem
cells in normal lung and lung cancer. Cell 121,
823–835 (2005).
This study was the first to characterize a bipotent
cell type in the distal lung, termed the BASC.
Importantly, the BASCs were shown to expand in
response to oncogenic KRAS activation, suggesting
that they could function as a cell of origin for lung
ADC.
57. Tiozzo, C. et al. Deletion of Pten expands lung
epithelial progenitor pools and confers resistance to
airway injury. Am. J. Respir. Crit. Care Med. 180,
701–712 (2009).
58. Ventura, J. J. et al. p38α MAP kinase is essential in
lung stem and progenitor cell proliferation and
differentiation. Nature Genet. 39, 750–758 (2007).
59. Xu, X. et al. Evidence for type II cells as cells of origin
of K‑Ras-induced distal lung adenocarcinoma. Proc.
Natl Acad. Sci. USA109, 4910–4915 (2012).
This study expanded our understanding of cells of
origin in GEMMs of lung ADC by using cell-type
specific promoters to drive Cre-Lox Kras activation.
In contrast to previous theories, the authors find
that AT2 cells, rather than BASCs or club cells, are
the most likely cells of origin for KRAS-driven
tumours.
60. Desai, T. J., Brownfield, D. G. & Krasnow, M. A.
Alveolar progenitor and stem cells in lung
development, renewal and cancer. Nature 507,
190–194 (2014).
61. Mainardi, S. et al. Identification of cancer initiating
cells in K‑Ras driven lung adenocarcinoma. Proc. Natl
Acad. Sci. USA 111, 255–260 (2014).
© 2014 Macmillan Publishers Limited. All rights reserved
62. Rowbotham, S. P. & Kim, C. F. Diverse cells at the
origin of lung adenocarcinoma. Proc. Natl Acad. Sci.
USA 111, 4745–4746 (2014).
63. Sutherland, K. D. et al. Multiple cells‑of‑origin of
mutant K‑Ras–induced mouse lung adenocarcinoma.
Proc. Natl Acad. Sci. USA 111, 4952–4957 (2014).
64. Giangreco, A. et al. β‑Catenin determines upper
airway progenitor cell fate and preinvasive squamous
lung cancer progression by modulating epithelial–
mesenchymal transition. J. Pathol. 226, 575–587
(2012).
65. Ji, H. et al. LKB1 modulates lung cancer
differentiation and metastasis. Nature 448, 807–810
(2007).
This study validated Lkb1 as a bona fide tumour
suppressor in NSCLC by deleting Lkb1 in the
KRAS-driven GEMM of lung cancer. Lkb1 loss not
only potentiated tumour development but also
promoted metastasis and drove tumours with
mixed ADC and SCC characteristics.
66. Malkoski, S. P. et al. Role of PTEN in basal cell derived
lung carcinogenesis. Mol. Carcinog. http://dx.doi.
org/10.1002/mc.22030 (2013).
67. Beck, B. & Blanpain, C. Unravelling cancer stem cell
potential. Nature Rev. Cancer 13, 727–738 (2013).
68. Li, F., Tiede, B., Massague, J. & Kang, Y. Beyond
tumorigenesis: cancer stem cells in metastasis. Cell
Res. 17, 3–14 (2007).
69. Jiang, F. et al. Aldehyde dehydrogenase 1 is a tumor
stem cell-associated marker in lung cancer. Mol.
Cancer Res. 7, 330–338 (2009).
70. Lau, A. N. et al. Tumor-propagating cells and Yap/Taz
activity contribute to lung tumor progression and
metastasis. EMBO J. 33, 468–481 (2014).
71. Liu, J. et al. Lung cancer tumorigenicity and drug
resistance are maintained through ALDHhiCD44hi
tumor initiating cells. Oncotarget 4, 1698–1711
(2013).
72. Sullivan, J. P. et al. Aldehyde dehydrogenase activity
selects for lung adenocarcinoma stem cells dependent
on Notch signaling. Cancer Res. 70, 9937–9948
(2010).
73. Zheng, Y. et al. A rare population of
CD24+ITGB4+Notchhi cells drives tumor propagation
in NSCLC and requires Notch3 for self-renewal. Cancer
Cell 24, 59–74 (2013).
74. Curtis, S. J. et al. Primary tumor genotype is an
important determinant in identification of lung cancer
propagating cells. Cell Stem Cell 7, 127–133 (2010).
In this study, an orthotopic serial transplantation
model of primary murine lung tumour cells was
used to identify the first bona fide TPC population
in lung cancer. Importantly, TPC phenotype was
shown to be dependent on primary tumour
genotype, suggesting that TPC characteristics will
differ for different driver mutations.
75. Sequist, L. V. et al. Genotypic and histological
evolution of lung cancers acquiring resistance to EGFR
inhibitors. Sci. Transl. Med. 3, 75ra26 (2011).
76. Song, H. et al. Functional characterization of
pulmonary neuroendocrine cells in lung development,
injury, and tumorigenesis. Proc. Natl Acad. Sci. USA
109, 17531–17536 (2012).
77. Park, K.‑S. et al. Characterization of the cell of origin
for small cell lung cancer. Cell Cycle 10, 2806–2815
(2011).
78. Sutherland, K. D. et al. Cell of origin of small cell lung
cancer: inactivation of Trp53 and Rb1 in distinct cell
types of adult mouse lung. Cancer Cell 19, 754–764
(2011).
79. Gazdar, A. F., Girard, L., Lockwood, W. W., Lam, W. L.
& Minna, J. D. Lung cancer cell lines as tools for
biomedical discovery and research. J. Natl Cancer
Inst. 102, 1310–1321 (2010).
80. Dow, L. E. & Lowe, S. W. Life in the fast lane:
mammalian disease models in the genomics era. Cell
148, 1099–1109 (2012).
81. Levy, M. A., Lovly, C. M. & Pao, W. Translating
genomic information into clinical medicine: lung cancer
as a paradigm. Genome Res. 22, 2101–2108 (2012).
82. Shaw, A. T., Hsu, P. P., Awad, M. M. & Engelman, J. A.
Tyrosine kinase gene rearrangements in epithelial
malignancies. Nature Rev. Cancer 13, 772–787
(2013).
83. Camidge, D. R. et al. Activity and safety of crizotinib in
patients with ALK-positive non-small-cell lung cancer:
updated results from a phase 1 study. Lancet Oncol.
13, 1011–1019 (2012).
84. Mok, T. S. et al. Gefitinib or carboplatin-paclitaxel in
pulmonary adenocarcinoma. N. Engl. J. Med. 361,
947–957 (2009).

85. Shaw, A. T. et al. Effect of crizotinib on overall survival
in patients with advanced non-small-cell lung cancer
harbouring ALK gene rearrangement: a retrospective
analysis. Lancet Oncol. 12, 1004–1012 (2011).
86. Sequist, L. V. et al. Phase I study of BGJ398, a
selective pan-FGFR inhibitor in genetically
preselected advanced solid tumors. AACR Meet.
[Abstr] CT326 (2014).
87. Katayama, R. et al. Mechanisms of acquired crizotinib
resistance in ALK-rearranged lung cancers. Sci. Transl.
Med. 4, 120ra17 (2012).
88. Kobayashi, S. et al. EGFR mutation and resistance of
non-small-cell lung cancer to gefitinib. N. Engl. J. Med.
352, 786–792 (2005).
89. Shaw, A. T. et al. Ceritinib in ALK-rearranged non-
small-cell lung cancer. N. Engl. J. Med. 370,
1189–1197 (2014).
90. Walter, A. O. et al. Discovery of a mutant-selective
covalent inhibitor of EGFR that overcomes
T790M‑mediated resistance in NSCLC. Cancer Discov.
3, 1404–1415 (2013).
91. Cross, D. A. et al. AZD9291, an irreversible EGFR TKI,
overcomes T790M‑mediated resistance to EGFR
inhibitors in lung cancer. Cancer Discov. http://dx.doi.
org/10.1158/2159-8290.CD-14-0337 (2014).
92. Yonesaka, K. et al. Activation of ERBB2 signaling
causes resistance to the EGFR-directed therapeutic
antibody cetuximab. Sci. Transl. Med. 3, 99ra86
(2011).
93. Zhang, Z. et al. Activation of the AXL kinase causes
resistance to EGFR-targeted therapy in lung cancer.
Nature Genet. 44, 852–860 (2012).
94. Ercan, D. et al. Reactivation of ERK signaling causes
resistance to EGFR kinase inhibitors. Cancer Discov. 2,
934–947 (2012).
95. Lin, L. & Bivona, T. G. Mechanisms of resistance to
epidermal growth factor receptor inhibitors and novel
therapeutic strategies to overcome resistance in
NSCLC patients. Chemother. Res. Pract. 2012,
817297 (2012).
96. Bivona, T. G. et al. FAS and NF‑κB signalling modulate
dependence of lung cancers on mutant EGFR. Nature
471, 523–526 (2011).
97. Xu, K. et al. EZH2 Oncogenic Activity in Castration-
Resistant Prostate Cancer Cells Is Polycomb-
Independent. Science 338, 1465–1469 (2012).
98. Chen, Z. et al. A murine lung cancer co‑clinical trial
identifies genetic modifiers of therapeutic response.
Nature 483, 613–617 (2012).
This study was the first of its kind to use GEMMs to
conduct a co-clinical trial that mirrored an ongoing
human clinical trial in patients with KRAS-mutant
lung cancer. The study found genetic predictors of
therapy response and provided a model for future
co-clinical trials aimed at accelerating the pace of
novel therapeutic development.
99. Engelman, J. A. et al. Effective use of PI3K and MEK
inhibitors to treat mutant Kras G12D and PIK3CA
H1047R murine lung cancers. Nature Med. 14,
1351–1356 (2008).
100. Janne, P. A. et al. Selumetinib plus docetaxel for
KRAS-mutant advanced non-small-cell lung cancer: a
randomised, multicentre, placebo-controlled, phase 2
study. Lancet Oncol. 14, 38–47 (2013).
101. Lim, S. M. et al. Therapeutic targeting of oncogenic
K‑Ras by a covalent catalytic site inhibitor. Angew.
Chem. Int. Ed Engl. 53, 199–204 (2014).
102. Ostrem, J. M., Peters, U., Sos, M. L., Wells, J. A. &
Shokat, K. M. K‑RasG12C inhibitors allosterically control
GTP affinity and effector interactions. Nature 503,
548–551 (2013).
103. Bass, A. J. et al. SOX2 is an amplified lineage-survival
oncogene in lung and esophageal squamous cell
carcinomas. Nature Genet. 41, 1238–1242 (2009).
104. Iwakawa, R. et al. MYC amplification as a prognostic
marker of early-stage lung adenocarcinoma identified
by whole genome copy number analysis. Clin. Cancer
Res. 17, 1481–1489 (2011).
105. Lin, C. Y. et al. Transcriptional amplification in tumor
cells with elevated c‑Myc. Cell 151, 56–67 (2012).
106. Engelman, J. A. & Janne, P. A. Mechanisms of acquired
resistance to epidermal growth factor receptor
tyrosine kinase inhibitors in non-small cell lung cancer.
Clin. Cancer Res. 14, 2895–2899 (2008).
107. Grosso, J. F. & Jure-Kunkel, M. N. CTLA‑4 blockade in
tumor models: an overview of preclinical and
translational research. Cancer Immun. 13, 5 (2013).
108. Lee, D. W., Barrett, D. M., Mackall, C., Orentas, R. &
Grupp, S. A. The future is now: chimeric antigen
receptors as new targeted therapies for childhood
cancer. Clin. Cancer Res. 18, 2780–2790 (2012).
NATURE REVIEWS | CANCER VOLUME 14 | AUGUST 2014 | 545
109. Tseng, D. et al. Anti‑CD47 antibody-mediated
phagocytosis of cancer by macrophages primes an
effective antitumor T‑cell response. Proc. Natl Acad.
Sci. USA 110, 11103–11108 (2013).
110. Zhang, B. CD73: a novel target for cancer
immunotherapy. Cancer Res. 70, 6407–6411
(2010).
111. Sandler, A. et al. Paclitaxel-carboplatin alone or with
bevacizumab for non-small-cell lung cancer. N. Engl.
J. Med. 355, 2542–2550 (2006).
112. Schoenfeld, J. et al. Active immunotherapy induces
antibody responses that target tumor angiogenesis.
Cancer Res. 70, 10150–10160 (2010).
113. Yuan, J. et al. Pretreatment serum VEGF is associated
with clinical response and overall survival in advanced
melanoma patients treated with ipilimumab. Cancer
Immunol. Res. 2, 127–132 (2014).
114. Lee, J.‑H. et al. Lung stem cell differentiation in mice
directed by endothelial cells via a
BMP4‑NFATc1‑thrombospondin‑1 axis. Cell 156,
440–455 (2014).
115. Straussman, R. et al. Tumour micro-environment
elicits innate resistance to RAF inhibitors through HGF
secretion. Nature 487, 500–504 (2012).
116. Filippakopoulos, P. et al. Selective inhibition of BET
bromodomains. Nature 468, 1067–1073 (2010).
117. Shimamura, T. et al. Efficacy of BET bromodomain
inhibition in Kras-mutant non-small cell lung cancer.
Clin. Cancer Res. 19, 6183–6192 (2013).
118. Liu, Y. et al. Metabolic and functional genomic studies
identify deoxythymidylate kinase as a target in
LKB1‑mutant lung cancer. Cancer Discov. 3, 870–879
(2013).
119. Molina-Arcas, M., Hancock, D. C., Sheridan, C.,
Kumar, M. S. & Downward, J. Coordinate direct input
of both KRAS and IGF1 receptor to activation of PI3
kinase in KRAS-mutant lung cancer. Cancer Discov. 3,
548–563 (2013).
120. Orita, H. et al. Selective inhibition of fatty acid
synthase for lung cancer treatment. Clin. Cancer Res.
13, 7139–7145 (2007).
121. Shackelford, D. B. et al. LKB1 inactivation dictates
therapeutic response of non-small cell lung cancer to
the metabolism drug phenformin. Cancer Cell 23,
143–158 (2013).
122. Hughes, L. A. et al. The CpG island methylator
phenotype: what’s in a name? Cancer Res. 73,
5858–5868 (2013).
123. Losman, J. A. & Kaelin, W. G. Jr. What a difference a
hydroxyl makes: mutant IDH, (R)-2‑hydroxyglutarate,
and cancer. Genes Dev. 27, 836–852 (2013).
124. Shinjo, K. et al. Integrated analysis of genetic and
epigenetic alterations reveals CpG island methylator
phenotype associated with distinct clinical characters
of lung adenocarcinoma. Carcinogenesis 33,
1277–1285 (2012).
125. Chen, Z. et al. Inhibition of ALK, PI3K/MEK, and
HSP90 in murine lung adenocarcinoma induced by
EML4‑ALK fusion oncogene. Cancer Res. 70,
9827–9836 (2010).
126. Ji, H. et al. The impact of human EGFR kinase domain
mutations on lung tumorigenesis and in vivo
sensitivity to EGFR-targeted therapies. Cancer Cell 9,
485–495 (2006).
127. Chen, Z. et al. Co‑clinical trials demonstrate superiority
of crizotinib to chemotherapy in ALK-rearranged non-
small cell lung cancer and predict strategies to overcome
resistance. Clin. Cancer Res. 20, 1204–1211 (2014).
128. Jackson, E. L. et al. The Differential effects of mutant
p53 alleles on advanced murine lung cancer. Cancer
Res. 65, 10280–10288 (2005).
129. Oliver, T. G. et al. Chronic cisplatin treatment
promotes enhanced damage repair and tumor
progression in a mouse model of lung cancer. Genes
Dev. 24, 837–852 (2010).
130. Blasco, R. B. et al. c‑Raf, but not B‑Raf, is essential for
development of K‑Ras oncogene-driven non-small cell
lung carcinoma. Cancer Cell 19, 652–663 (2011).
131. Puyol, M. et al. A synthetic lethal interaction between
K‑Ras oncogenes and Cdk4 unveils a therapeutic
strategy for non-small cell lung carcinoma. Cancer Cell
18, 63–73 (2010).
132. Savino, M. et al. The action mechanism of the Myc
inhibitor termed Omomyc may give clues on how to
target Myc for cancer therapy. PLoS ONE 6, e22284
(2011).
133. Soucek, L. et al. Modelling Myc inhibition as a cancer
therapy. Nature 455, 679–683 (2008).
134. Soucek, L. et al. Inhibition of Myc family proteins
eradicates KRas-driven lung cancer in mice. Genes
Dev. 27, 504–513 (2013).
© 2014 Macmillan Publishers Limited. All rights reserved
REVIEWS
135. Zhang, W. C. et al. Glycine decarboxylase activity
drives non-small cell lung cancer tumor-initiating cells
and tumorigenesis. Cell 148, 259–272 (2012).
136. Rock, J. R. et al. Basal cells as stem cells of the mouse
trachea and human airway epithelium. Proc. Natl
Acad. Sci. USA 106, 12771–12775 (2009).
In this study, the KRT5+ and p63+ basal cells of the
trachea were shown to be true stem or progenitor
cells through lineage tracing experiments during
development and tracheal epithelial repair. The
authors isolated similar cells from human tracheas
and showed that both murine and human cells can
be cultured as tracheospheres in three-dimensional
Matrigel-containing transwells.
137. Hegab, A. E. et al. Novel stem/progenitor cell
population from murine tracheal submucosal gland
ducts with multipotent regenerative potential.
Stem Cells 29, 1283–1293 (2011).
138. Rock, J. R. & Hogan, B. L. M. Epithelial progenitor cells
in lung development, maintenance, repair, and disease.
Annu. Rev. Cell Dev. Biol. 27, 493–512 (2011).
139. Rawlins, E. L. et al. The role of Scgb1a1+ Clara cells in
the long-term maintenance and repair of lung airway,
but not alveolar, epithelium. Cell Stem Cell 4,
525–534 (2009).
140. Teixeira, V. H. et al. Stochastic homeostasis in human
airway epithelium is achieved by neutral competition of
basal cell progenitors. eLife 2, http://dx.doi.org/
10.7554/eLife.00966#sthash.XxrCqaIk.dpuf (2013).
141. Teisanu, R. M., Lagasse, E., Whitesides, J. F. &
Stripp, B. R. Prospective Isolation of Bronchiolar Stem
Cells Based Upon Immunophenotypic and
Autofluorescence Characteristics. Stem Cells 27,
612–622 (2009).
142. Barkauskas, C. E. et al. Type 2 alveolar cells are stem
cells in adult lung. J. Clin. Invest. 123, 3025–3036
(2013).
This study was the first to show that AT2 cells can
function as facultative stem or progenitor cells in
the adult lung to differentiate into both AT2 and
AT1 lineages after injury.
143. Chapman, H. A. et al. Integrin α6β4 identifies an adult
distal lung epithelial population with regenerative
potential in mice. J. Clin. Invest. 121, 2855–2862
(2011).
144. Tropea, K. A. et al. Bronchioalveolar stem cells increase
after mesenchymal stromal cell treatment in a mouse
model of bronchopulmonary dysplasia. Am. J. Physiol.
Lung Cell. Mol. Physiol. 302, L829–L837 (2012).
145. Zacharek, S. J. et al. Lung stem cell self-renewal relies
on BMI1‑dependent control of expression at imprinted
loci. Cell Stem Cell 9, 272–281 (2010).
146. Rock, J. R. et al. Multiple stromal populations
contribute to pulmonary fibrosis without evidence for
epithelial to mesenchymal transition. Proc. Natl Acad.
Sci. USA 108, E1475–E1483 (2011).
147. McQualter, J. L., Yuen, K., Williams, B. &
Bertoncello, I. Evidence of an epithelial stem/
progenitor cell hierarchy in the adult mouse lung.
Proc. Natl Acad. Sci. USA 107, 1414–1419 (2010).
148. Hegab, A. E. et al. Isolation and in vitro
characterization of basal and submucosal gland duct
stem/progenitor cells from human proximal airways.
Stem Cells Translat. Med. 1, 719–724 (2012).
149. Kajstura, J. et al. Evidence for human lung stem cells.
New Engl. J. Med. 364, 1795–1806 (2011).
150. Marek, L. et al. Fibroblast growth factor (FGF) and
FGF receptor-mediated autocrine signaling in non-
small-cell lung cancer cells. Mol. Pharmacol. 75,
196–207 (2009).
151. Liao, R. G. et al. Inhibitor-sensitive FGFR2 and FGFR3
mutations in lung squamous cell carcinoma. Cancer
Res. 73, 5195–5205 (2013).
152. Xu, L. et al. Combined EGFR/MET or EGFR/HSP90
inhibition is effective in the treatment of lung cancers
codriven by mutant EGFR containing T790M and MET.
Cancer Res. 72, 3302–3311 (2012).
153. Davies, K. D. et al. Identifying and targeting ROS1
gene fusions in non-small cell lung cancer. Clin. Cancer
Res. 18, 4570–4579 (2012).
154. Takeuchi, K. et al. RET, ROS1 and ALK fusions in lung
cancer. Nature Med. 18, 378–381 (2012).
155. Wang, S. E. et al. HER2 kinase domain mutation
results in constitutive phosphorylation and activation
of HER2 and EGFR and resistance to EGFR tyrosine
kinase inhibitors. Cancer Cell 10, 25–38 (2006).
156. McDermott, U. et al. Ligand-dependent platelet-
derived growth factor receptor (PDGFR)-α activation
sensitizes rare lung cancer and sarcoma cells to
PDGFR kinase inhibitors. Cancer Res. 69,
3937–3946 (2009).
REVIEWS
157. Jackson, E. L. et al. Analysis of lung tumor initiation
and progression using conditional expression of
oncogenic K‑ras. Genes Dev. 15, 3243–3248 (2001).
GEMMs of lung cancer have been developed largely
based on this prototypical study, in which the
Cre-Lox system was used to induce oncogenic Kras
expression specifically in the lung through
Adeno-Cre intranasal instillation.
158. Mascaux, C. et al. The role of RAS oncogene in
survival of patients with lung cancer: a systematic
review of the literature with meta-analysis. Br.
J. Cancer 92, 131–139 (2005).
159. De Raedt, T. et al. Exploiting cancer cell vulnerabilities
to develop a combination therapy for ras-driven
tumors. Cancer Cell 20, 400–413 (2011).
160. Paik, P. K. et al. Clinical characteristics of patients with
lung adenocarcinomas harboring BRAF mutations.
J. Clin. Oncol. 29, 2046–2051 (2011).
161. Kawano, O. et al. PIK3CA mutation status in Japanese
lung cancer patients. Lung Cancer 54, 209–215
(2006).
162. Marks, J. L. et al. Novel MEK1 mutation identified by
mutational analysis of epidermal growth factor receptor
signaling pathway genes in lung adenocarcinoma.
Cancer Res. 68, 5524–5528 (2008).
163. Licciulli, S. et al. Notch1 is required for Kras-induced
lung adenocarcinoma and controls tumor cell survival
via p53. Cancer Res. 73, 5974–5984 (2013).
164. Westhoff, B. et al. Alterations of the Notch pathway in
lung cancer. Proc. Natl Acad. Sci. USA 106,
22293–22298 (2009).
546 | AUGUST 2014 | VOLUME 14 www.nature.com/reviews/cancer
165. Issaeva, I. et al. Knockdown of ALR (MLL2) reveals
ALR target genes and leads to alterations in cell
adhesion and growth. Mol. Cell. Biol. 27,
1889–1903 (2007).
166. Takawa, M. et al. Validation of the histone
methyltransferase EZH2 as a therapeutic target for
various types of human cancer and as a prognostic
marker. Cancer Sci. 102, 1298–1305 (2011).
167. Yang, H. et al. Tumor development is associated with
decrease of TET gene expression and 5‑methylcytosine
hydroxylation. Oncogene 32, 663–669 (2013).
168. Gao, Q. et al. Deletion of the de novo DNA
methyltransferase Dnmt3a promotes lung tumor
progression. Proc. Natl Acad. Sci. USA 108,
18061–18066 (2011).
169. Ohta, T. et al. Loss of Keap1 function activates Nrf2
and provides advantages for lung cancer cell growth.
Cancer Res. 68, 1303–1309 (2008).
170. Singh, A. et al. Dysfunctional KEAP1‑NRF2 interaction
in non-small-cell lung cancer. PLoS Med. 3, e420
(2006).
171. Wang, Y., Zhang, Z., Kastens, E., Lubet, R. A. &
You, M. Mice with alterations in both p53 and Ink4a/
Arf display a striking increase in lung tumor multiplicity
and progression: differential chemopreventive effect of
budesonide in wild-type and mutant A/J mice. Cancer
Res. 63, 4389–4395 (2003).
172. Lou-Qian, Z. et al. The prognostic value of epigenetic
silencing of p16 gene in NSCLC patients: a systematic
review and meta-analysis. PLoS ONE 8, e54970
(2013).
© 2014 Macmillan Publishers Limited. All rights reserved
173. Mitsudomi, T., Hamajima, N., Ogawa, M. &
Takahashi, T. Prognostic significance of p53
alterations in patients with non-small cell lung cancer:
a meta-analysis. Clin. Cancer Res. 6, 4055–4063
(2000).
174. Sanchez-Cespedes, M. et al. Inactivation of LKB1/
STK11 is a common event in adenocarcinomas of the
lung. Cancer Res. 62, 3659–3662 (2002).
175. Iwanaga, K. et al. Pten inactivation accelerates
oncogenic K‑ras-initiated tumorigenesis in a mouse
model of lung cancer. Cancer Res. 68, 1119–1127
(2008).
176. Soria, J. C. et al. Lack of PTEN expression in non-
small cell lung cancer could be related to promoter
methylation. Clin. Cancer Res. 8, 1178–1184
(2002).
Acknowledgements
The authors thank United Against Lung Cancer, Thoracic
Foundation, Bonnie J Addario Lung Cancer Foundation,
Claudia Adams Barr Program For Basic Cancer Research,
grant numbers CA122794, CA166480, CA163896,
CA154303, CA120964 CA140594.
Competing interests statement
The authors declare no competing interests.
DATABASES
My Cancer Genome: http://www.mycancergenome.org/
ALL LINKS ARE ACTIVE IN THE ONLINE PDF